What is wrong with op-amps?

jean-paul · 2010-06-29 2:09 pm

There is a brand new ipod at my desk and it is winking at me for months now.

Sebastiaan · 2010-06-29 2:10 pm

Opamps, Tubes, FET's Transistors or an combination of all. I don't care. If it sounds good it is good!

jean-paul · 2010-06-29 2:19 pm

You forgot the ubiquitous :"and if it measures good" part...

Good sounding or pleasant sounding might as well be a good part distortion. The ears are good instruments but checking with measuring equipment is better.

Alan Frobisher · 2010-06-29 2:55 pm

I concur with Mooly,had some great success with OPA604(single)/2604(dual).Operate with good headroom (rail voltage) and roll off the freq.response with a small cap.across the fb.resistor at 3-5 times required freq.

suntechnik · 2010-06-29 3:16 pm

wahab said:
musically speaking :

I am not a musian and apologizing the way I quoted the message in case it was done wrong. I assumed since the original message was above I can emphasis just a few words not repost everything again.

OK to be clear (sorry english is 2-nd foring language for me) my point is that studing high math I never heard that Fourier proved the odd harmonic sounds better then even. I could assume only that some guys were feeding different audio components with 1KHz test signal and while studying the spectrum coming out made an assumption later that AUDIBLY components having prevailing odd harmonics distortion as not as bad sounding as components having prevailing even harmonics distortion (at 1KHz). The validity of such statement might be considered at the following : "Tiramisu is a more delicious desert then an apple pie" or "NFB sounds bad".

However clean sounding device means 1KHz in ONLY 1KHz out. I would say Balanced / differential signal path has major impact here not tube vs transistor IMHO.

tiefbassuebertr · 2010-06-29 3:55 pm

jean-paul said:
This is also the way I see it except that I would like to add OPA627 and LME49860 (a dual opamp) to the list of good devices. Since most of the commercial gear uses dual opamps the single versions are useless unless one wants to use pink cat adapter PCB's etc. which can make things worse.

As usual a non-technical opinion from me: in all cases where I replaced already tweaked opamp circuits (with even very good types) for an open loop discrete design with all its disadvantages it still sounded better. And no I did not listen to added distortion. The standard disadvantages with discrete circuits are :high offset/drift, channel differences etc. in a way larger magnitude than with opamps.

My experiences are with I/V converters in DACs and preamps. But it often is poor implementation of the opamps in question that defines the results. Proper decoupling the power supply lines is such an example where things go wrong.

Opamps are quite easy to work with when you know their limitations but a simple discrete circuit often sounds better but only so after a lot more work.

Despite all excellent explanations from true experts I simplified matters by avoiding designs around opamps as I think the feedback is what causes the disadvantages.

Yeah, "I think" is not the best of reasoning. And I too leave opamp circuits sometimes as they are and replace the devices for better types and decouple them as it is far less work than redesigning. The point is to think carefully especially with high speed high bandwidth opamps. An input filter around 100 khz can be a good remedy when it picks up high frequency signals.

BTW the original power supply of NAD PP-2 is an example of how not to do it !!

Regarded the OPA627 I can not understand why some count him as to the top class operational amps. The OPA604 and OPA134 I would prefer both for high gain stage (RIAA) and low gain stage (line).
In case of I/U converter there is a easy way to find out the sonic character of various op amps:
compare it to an easy resistor load like showed here :
Digi Scoop 1
until this day I haven't find out an OP Amp, that clearly outperform a simple resistor. For me is the best I/U converter that one from the Pass "D1" (very easy, but at my view very very hard to beat).

If more than one voltage gain stages is in the negative feedback loop, causes this the main disadvantages together with too low idle current in the power buffer stage. This means, not the feedback itself is the reason.

Mooly notice, that the OPA604/2604 sounds musical. Also one of my favorite. Additional the AD797 (Mr. Scott Wurzer's Design) and the OPA134/2134.

If you study the internal circuit, you will understand, why; only one voltage gain stage there is in the NFB loop and enough current through the last stage. This is for me fundamental for best sounding OP-Amps (and best sounding discrete line/power audio amps). Two gain stages in the NFB loop (still present by the most OP amp types) creates still high/odd order harmonics and thus harsh sound.

Interesting would be to know the internal topology from the LME series (and that one from newest TI/Analog Devices types). But I think, there is also only one voltage gain stage in the NFB loop - some guy's think, same basic topology than AD797.

BTW - I have compare some years ago two unity gain line preamp concepts, one with OPA627 and one with Andrea Ciuffoli's power follower in modified preamp line version (only 9V but 1A idle current). Unfortunately I haven't documentation of the measurement results. But it was to observe follow: At 1 KHz and 10 KHz much more better THD results by the OPA627. And by 100 KHz and 500 KHz and soundcheck clearly better results by the power follower ( OPA627 deliver sathtooth instead sine wave by 500 KHz). 2xIRF520 was in use by the power follower. Sonic quality was better in all respects by Ciuffoli's power follower. The used power amp here was a ZEN diy project and a tube power amp from VTL (see attachement). And the speakers at this time was a 2-way combination of PHL B17-1220 and Focal T120K (very high resolution loudspeakers);
have a look also to this this URL:
http://www.diyaudio.com/forums/pass...itional-op-amp-ultimate-sounding-phl1230.html
The also still present preamp VTL "Maximal" was in the perception between both, clearly better than the preamp with OPA627, clearly inferior to the power follower concept.

bob91343 · 2010-06-29 6:54 pm

TI has a new op amp line that looks great; I'm getting some. They are JFET input, very low bias and offset and static current, good bandwidth (11 MHz) and noise, and CHEAP. Intended for high end audio. OPA1641.

tiefbassuebertr · 2010-06-29 7:52 pm

jean-paul · 2010-06-29 8:08 pm

Hi, just like CarlosFM did soem years ago I discovered OPA627 requires careful decoupling with an extra cap across the power pins. I still like it very much.

Testing at 500 kHz does not say much to me as I don't have any music that reaches such frequencies

AD797 is a difficult one to control/tame. I only use it in Flea power supplies.

MatchASM · 2010-06-29 8:51 pm

teemuk said:
There are just awful alot of myths and erroneous beliefs that are reigning in the scene of audio electronics. The belief that discrete circuits for some mystical reason are better than integrated circuits is one of them.

There is nothing mystical about capacitive and inductive coupling of circuit nodes inside a standard IC package. They put a limit on the GBW that an opamp can have without becoming unstable. It's difficult to simulate accurately, that's all.

Discretes, if designed and executed well, don't have this limitation.

jan.didden · 2010-06-29 9:39 pm

MatchASM said:
There is nothing mystical about capacitive and inductive coupling of circuit nodes inside a standard IC package. They put a limit on the GBW that an opamp can have without becoming unstable. It's difficult to simulate accurately, that's all.

Discretes, if designed and executed well, don't have this limitation.

Are you sure about this? There are opamps with 2.5GHz GBW that are absolutely stable. I don't think any discrete circuit can get close to that within a lightyear. Did I miss something?

jd

jean-paul · 2010-06-29 9:43 pm

I think it is in the "standard IC package"....

tpsorin · 2010-06-29 10:09 pm

There is nothing wrong with opamps. All music that is recorded goes tru a ton of opamps in studio.
What is the point in worrying for having one/two more opamp in signal chain?

MatchASM · 2010-06-29 10:36 pm

janneman said:
There are opamps with 2.5GHz GBW that are absolutely stable. I don't think any discrete circuit can get close to that within a lightyear.

Maybe we differ in our interpretation of what constitutes a "discrete" amplifier. I'm pretty sure (according to what I consider discrete) there are discrete microwave amplifiers that have GBWs that are massively larger than any "standard IC package" opamp. The package is, for example, the reason that you will never find a Ku-band (12GHz) amplifier with more than 40dB gain in a standard IC package. I consider cascading a couple of such amplifiers on a PCB a "discrete" solution.

benb · 2010-06-29 11:51 pm

wahab said:
What is wrong is that some people think that their
ears are accurate enough to measure -100db THD ratios
as soon as an op amp is involved in the sound chain..

Yet, a tube amp that produce 2% of THD will sound
distorsionless for the same pundits..

tiefbassuebertr said:
Yes, and this is still again to observe in a hifi shop by demonstrations of commercial preamp devices, if classic and acoustical/vocal jazz CD's/records was used. But what is the actually reason therefore?

janneman said:
The reason I don't know, but the cause is in the way perception works.

jd

Firstly, THD is a mostly bogus number, and is an unfortunate part of the ongoing tube-vs-transistor debate. Mr. Geddes (who I see posts elsewhere on diyaudio) has written excellent explanations on the audibility of different types of distortion, and has come up with better metrics of distortion that correspond much more closely to hearing than the traditional THD figure:
Perception

Distortion that consists only or mostly of lower harmonics (as generated by the slight and smooth curve of a tube transfer characteristic) is much less audible than that that with higher or wide-ranging harmonics (as in transistor circuits, especially crossover distortion where one transistor turns off and another turns on). The reason is psychoacoustic masking:
Auditory masking - Wikipedia, the free encyclopedia
In the case of lower (say, second, third and fourth) harmonics, they are relatively near the fundamental, and can be of significant amplitude before being heard. Higher harmonics, being further in frequency from the fundamental, are more easily heard at much lower levels than the fundamental.

Furthermore, I understand that negative feedback, while lowering distortion overall (giving a lower figure for the traditional THD measurement), spreads the distortion products around among the harmonics and thus can make distortion more audible (I've only heard this claim once, it seems like it would make sense, but I'm not sure - a little work with LTSpice should clear it up).

suntechnik said:
I am not a musian and apologizing the way I quoted the message in case it was done wrong. I assumed since the original message was above I can emphasis just a few words not repost everything again.

OK to be clear (sorry english is 2-nd foring language for me) my point is that studing high math I never heard that Fourier proved the odd harmonic sounds better then even. I could assume only that some guys were feeding different audio components with 1KHz test signal and while studying the spectrum coming out made an assumption later that AUDIBLY components having prevailing odd harmonics distortion as not as bad sounding as components having prevailing even harmonics distortion (at 1KHz). The validity of such statement might be considered at the following : "Tiramisu is a more delicious desert then an apple pie" or "NFB sounds bad".

However clean sounding device means 1KHz in ONLY 1KHz out. I would say Balanced / differential signal path has major impact here not tube vs transistor IMHO.

When equipment comes available with no distortion (neither measurable nor audible) at anything near affordable prices, we'll all stand in line for it like Apple fans buying the latest iPhone. But until then we'll be debating, and hopefully learning more about, these things.

It wasn't Fourier's spectrum analysis math that taught us about even vs. odd harmonics, that's just a tool that became applicable in recent decades to easily see the spectrum of a digitized waveform based on modern computing power. The earliest work goes back to Pythagoras who recognized the mathematical relationships between different musical tones:
Standing Waves

rickedd46 · 2010-06-30 12:52 am

Are people as critical of the topology implemented in the recording equipment as they are with their personal equipment?

Is there a way to find out which particular topology was used for a particular recording either if there were tubes, discrete or op-amps used?

If you prefer tubes or transistors would you still listen to a recording where the recording equipment is stuffed with op-amps?

Can anyone pinpoint a recording made with op-amps, tubes or transistors just by listening to it?

Are all recordings considered perfect regardless of topology?

Sorry for so many questions. I am totally ignorant to a lot of this stuff.

Ricardo.

indianajo · 2010-06-30 1:21 am

tube vs op amp playoff

Well, if anybody cares to listen to a non-scientist whose only instruments are a scope, a dvm, and some nice speakers, I've been running a playoff of the all tube Dynakit PAS2 vs it's "replacement", an dual op amp based Herald RA-88a "disco mixer", for amplifying RIAA records and a CD player. The bottom line is, details matter.
I had trouble with the PAS2, and it's first "replacement", a Sony TC250 tape recorder, and both had developed low volume on one channel. The sony had about 100 electrolytic capacitors and a really noisy line-record player switch, so I abandoned it and bought the disco mixer for $15. Balance problem solved, the problem wasn't in the turntable. But the disco mixer, with it's JRC 4558 op amps, had an annoying level of hiss. So I looked up "low noise" op amps on the Newark selection chart, and bought some ST33078 op amps. Instant gratification- much less noise. But the disco mixer now had an annoying level of hum on the record player, maybe 40 db down. Grounding the case to the 3rd pin safety ground, as well as the record player green wire, didn't help.
After retirement, I tried swapping the tubes of the PAS2, the problem stayed on one channel. I lifted one leg of each resistor on the RIAA board to measure it, all were fine. I reheated th joints. No improvement. So I bought some .1 polyprophylene capacitors and some smaller polystyrene ones, and replaced the paper capacitors on the phono board. One had a burn spot in the wax case from the builder in 1961. So, problem solved, the channels were balanced. I then replaced one carbon comp plate resistor with a metal film. Instant reduction of hiss by a db or two. Now the PAS 2 sounded better than the disco-mixer.
Not stopping there, I tried replacing the caps on the high level board of the PAS2 with polyprophylene. Instant disaster-galloping 20v transients on the B+ supply all across the amp. Pulling tubes doesn't help. Putting the paper caps back doesn't help. So right now, the 33078 op amp disco mixer sounds best. I just received a new 20-20-20-20 uf capacitor to replace the 1978 manufactured (purchased 1982) b+ electrolytic, to see if turning it upside down caused some water flow and a chemical problem.
In the same order, I received some isolated ground RCA jacks for the op amp disco mixer. I noticed that the hum free PAS2 has isolated ground on the magnetic phono input, the disco mixer grounds all RCA jack outsides to case, although using the negative input of the op amp to reject hum is some of the point of using an op amp. They did hook up the input of the op amp and provide it with expensive RIAA roll off capacitors, so really omiting this detail is the sign of a poorly managed company whose products sell for $15 at the flea market. Future hum performance results TBD.
On the subject of discretes vs op amps, my ST120 has 6 transistors per channel, plus 5 on the DJoffe mod devoted to calculating and supplying adequate output transistor bias current. My newish CS800S packaged PA amp has 4560 op amps on the input board, driving discrete phase splitters, pre-drivers, and output bipolard transistors. Adjusting the gain of the CS800s for living room volume, I can't really tell any difference between the two amps through the SP2 speakers. I also have an all tube Dynakit ST70, which is retired for 25 year old B+ capacitors and decade old output tubes. It is so expensive to keep a tube amp performing up to standard. I recall putting the ST70 in the attic, it was sounding tired on loud piano hits at the volume necessary to back up a church choir. I hope in the future to get it up to snuff with Sovtek tubes and a new capacitor and run it directly A-B-C vs the ST120 DJoffe, and the CS800S.

john curl · 2010-06-30 1:23 am

Many times we can. This is why many of the REALLY GOOD recordings were originally made in the 1950's and 1960's. Like 'Kind of Blue' or ' Belafonte at Carnegie Hall'. These recordings on vinyl are GREAT, with really good audio reproduction equipment, yet they were recorded with tubes over 50 years ago.

abraxalito · 2010-06-30 2:24 am

benb said:
Firstly, THD is a mostly bogus number, and is an unfortunate part of the ongoing tube-vs-transistor debate.

Its only bogus to the degree that the harmonics are all summed together to give a single number whereas its widely accepted that higher order ones are most offensive and least masked.

Mr. Geddes (who I see posts elsewhere on diyaudio) has written excellent explanations on the audibility of different types of distortion, and has come up with better metrics of distortion that correspond much more closely to hearing than the traditional THD figure:

Have you read his papers? I had a look and one thing I can't fathom is why he focusses almost exclusively on static distortion - distortion that is frequency independent and can be modelled by a bent transfer function. As far as I'm aware, that focus precludes examination of most of the known distortion mechanisms in audio amplifiers as outlined by Doug Self. As you're someone who seems to regard this work highly, do you have any idea of why he'd ignore dynamic distortion?

Furthermore, I understand that negative feedback, while lowering distortion overall (giving a lower figure for the traditional THD measurement), spreads the distortion products around among the harmonics and thus can make distortion more audible (I've only heard this claim once, it seems like it would make sense, but I'm not sure - a little work with LTSpice should clear it up).

I believe such work has been done - it shows that for modest levels of feedback this is indeed the case. However as the feedback is increased further the higher order components drop down monotonically with increasing NFB. So as Bruno Putzeys says 'if you're going to use feedback, use lots of it'.

wakibaki · 2010-06-30 2:52 am

Gordy said:
(If you say ‘they don’t sound good’ please give a technical reason why you think that might be).

As they say, opinions are like 4ssholes, everybody got one. Everybody likes to express theirs in a thread of this type. Everybody likes to ignore the fact that they are valueless, even though many should, and many do know better.

Equipment can be tested, using 'before and after' subtraction/comparison test gear. Equipment can be evaluated by so-called 'blind testing', the results of both types of tests should agree.

Almost all the results you will see presented here will be purely anecdotal. They will not be based on any testing of any kind. They will either be based on some kind of 'design philosophy' article of faith, supported by the subjective opinion of one person (or perhaps more) in uncontrolled conditions, or they will be based on the subjective opinion of one person (or perhaps more) in uncontrolled conditions, supported by some kind of ex-post-facto rationalisation. If there is one supported by subtractive testing, I will be shocked. Like I say, valueless.

You could print them all out, tear them up and assemble them into a large papier-mache opamp (non-functional, of course), call it 'hommage a diyaudio', and win the Turner prize, thereby confirming that they are in fact, valueless.

w

What is wrong with op-amps?

Disabled Account

Member

Disabled Account

Member

Member

Member

Attachments

Member

Member

Disabled Account

Member

AX tech editor

Disabled Account

Member

Member

Member

Member

Member

Member

Member

Banned